JP4817783B2 - RAID system and rebuild / copyback processing method thereof - Google Patents

Abstract

A RAID system access a physical disk according to a host I/O request, and perform Rebuild/Copy back processing, for implementing high-speed Rebuild/Copy back processing without interrupting the processing of a normal I/O. When one disk device out of a plurality of disk devices constituting the RAID configuration fails, the processing size for one time of Rebuild/Copy back to reconstruct the RAID is changed according to the presence of a normal I/O managed in RAID control. When a normal I/O is present, Rebuild/Copy back can be executed without interrupting the processing of the normal I/O, and in a static state where a normal I/O is absent, a processing size larger than the size in that state can be specified so as to implement Rebuild/Copy back processing at high-speed.

Description

  The present invention relates to a RAID system that stores data in a plurality of disk devices with redundancy, and rebuilds and copies that rebuild a redundant configuration with data from other disk devices when some disk devices fail More particularly, the present invention relates to a RAID system that performs rebuilding and copyback while receiving host I / O, and a rebuilding / copyback processing method thereof.

  In recent years, as various data has been digitized and handled on computers, data storage that can store large amounts of data efficiently and with high reliability, independent of the host computer that executes data processing The importance of devices (external storage devices) is increasing.

  As this data storage device, a disk array device composed of a large number of disk devices (for example, a magnetic disk device and an optical disk device) and a disk controller that controls these large number of disk devices is used. This disk array device employs a RAID configuration to realize data redundancy and improve reliability.

  In such a disk array device, when a disk device constituting a RAID group fails and loses redundancy, redundancy recovery is required. FIG. 8 is an explanatory diagram of a rebuild function of RAID 5 for such redundancy recovery. In order to perform active maintenance, spare disk devices HS (Hot Spare Disk) are provided in the four disk devices # 0, # 1, # 2, and # 3 constituting RAID5.

  The disk device group 160 is connected to a pair of disk controllers 110 and 120. Each of the disk controllers 110 and 120 includes a disk adapter 140 that controls an interface with the disk device group 160, a control unit 120, and a channel adapter 100 that controls an interface with a host (not shown).

  Here, when the disk device # 0 fails among the four disk devices constituting the RAID 5, the data of the disk devices # 1, # 2, and # 3 other than the failed disk device # 0 are transferred to the disk adapter. The data is read into a cache memory (not shown) of the control unit 120 via 140, and these XOR operations are performed to create redundant data.

  Then, the created redundant data is written to the spare disk device HS via the disk adapter 140 to restore redundancy. This is called the Rebuild function.

  On the other hand, FIG. 9 is an explanatory diagram of the copyback function. When the failed disk device # 0 is replaced with a new disk device New from the state where Rebuild is completed in FIG. 8, copy back processing is performed. In other words, this is a process of writing the redundant data written in the spare disk device HS back to the original disk device New.

It is desirable to execute such Rebuild / Copyback processing while processing I / O requests from the host, and a method for balancing the number of requests has been proposed (for example, Patent Document 1).
Japanese Patent Laying-Open No. 2004-295860 (FIG. 14)

  As described above, in order to perform Rebuild / Copyback processing while processing an I / O request from the host, processing for one disk device cannot be performed at a time. For this reason, the number of times of the disk device data amount is determined by fixing the processing size of one time, and reading the data from the normal disk device and writing the data to the write destination disk device for the fixed processing size. Was running.

  For example, in Rebuild, data is read from a normal disk device, redundant data other than RAID1 (mirroring) is created and written to a spare disk device HS or a new disk device New, and in Copyback, a spare disk is created. Data is read from the HS and written to the new disk device New.

  Conventionally, for these operations, the processing size for each operation is fixed and does not vary depending on the load state of the RAID group. Therefore, when there is normal I / O (access from the host), that is, in a state where there is some load, adjustment is made to balance disk access by normal I / O and disk access by Rebuild / Copyback. On the other hand, in the system, the performance of the disk device cannot be sufficiently brought out by the Rebuild / Copyback process when there is no normal I / O.

  For this reason, when there is no normal I / O, the time until completion of Rebuild / Copyback becomes longer compared to the performance of a disk device that is commonly considered. In recent years, since the storage capacity of disk devices has increased, it has become a problem to shorten the time required to complete Rebuild / Copyback.

  Therefore, an object of the present invention is to provide a RAID system for improving the performance of rebuild / copyback processing and rebuild / copyback processing thereof when balancing rebuild / copyback while processing host I / O requests. It is to provide a method.

  Another object of the present invention is to provide a RAID system and its rebuild / copy for shortening the rebuild / copyback completion time when balancing rebuild / copyback while processing host I / O requests. It is to provide a back processing method.

  Furthermore, another object of the present invention is to provide a RAID system for balancing rebuild / copyback while processing host I / O requests while shortening the completion time of rebuild / copyback, and its rebuild / copyback. It is to provide a processing method.

In order to achieve this object, the present invention accesses a connected disk device in response to an I / O request requested from a host and executes a plurality of I / O requests in a RAID system. When the disk device and one of the plurality of disk devices fail, the operation of reading the data of the disk device other than the failed disk device and writing the data to the spare disk device or the new disk device is divided into multiple times. The controller executes a rebuild or copyback process for reconstructing the RAID configuration, and the controller gives priority to the normal I / O request and the one request for the rebuild / copyback. Depending on the RAID control module to be executed at a predetermined ratio and the completion of the rebuild / copyback request processing, A rebuild / copyback module that issues a next rebuild / copyback request with reference to a progress table that manages the progress of the rebuild / copyback process, and the RAID control module is configured to send a normal I / O request The processing completion time is recorded every time processing of the normal I / O request is completed, and the processing completion time is compared with the current time to determine whether the normal I / O request comes within a predetermined time. The rebuild / copy back module inquires the determination result to the RAID control module in response to the completion of the rebuild / copy back request processing , and sets the request processing size for one rebuild or copy back processing as a normal I / O. If the request does not arrive within a predetermined time, set a larger value than when the normal I / O request arrives within a predetermined time. Once the request of rebuild / copy back, issued to the RAID control module.

The RAID system rebuild / copyback processing method according to the present invention reads the data of a disk device other than the failed disk device when one of the plurality of disk devices constituting the RAID fails, and a spare disk device. Or, the operation of writing data to the new disk device is executed in multiple steps, the rebuild or copyback process for rebuilding the RAID configuration is executed, and the normal I / O request processing is completed by the RAID control module The time is recorded every time processing of the normal I / O request is completed, the processing completion time is compared with the current time, and the normal I / O request including the I / O request from the host is within a predetermined time. A step of determining whether or not to come, and a single request for the normal I / O request and the rebuild / copyback by the RAID control module Are executed at a predetermined ratio in accordance with the priority, and the execution step of the rebuild / copy back process is performed by the rebuild / copy back module according to the completion of the rebuild / copy back request process. A step of starting generation of the next rebuild / copyback request with reference to a progress table managing the progress status of the rebuild / copyback processing, and the rebuild / copyback module processing the rebuild / copyback request Upon completion, the RAID control module is inquired about the determination result, and the request processing size of the rebuild or copyback process is set to one time when the normal I / O request does not arrive within a predetermined time. / O request is set larger than when it arrives within a predetermined time, and the rebuild / copy Once it requests for click, and a step of issuing to the RAID control module.

  In the present invention, it is preferable that the controller records the processing completion time of the normal I / O request every time processing of the normal I / O request is completed, and compares the processing completion time with the current time. Then, it is determined whether a normal I / O request has arrived for a predetermined time.

  In the present invention, it is preferable that the controller balances and executes the normal I / O request and one request of the rebuild / copyback and completes the processing of the rebuild / copyback request. In response, the next rebuild / copy back request is issued.

  In the present invention, it is preferable that the controller has a progress table for managing a progress status of the rebuild / copy back process, and creates a request for a rebuild or copy back process according to the progress table, The progress table is updated.

  In the present invention, it is preferable that the controller balances the normal I / O request and the one request for rebuild / copyback and executes the RAID control module and the rebuild / copyback request. A rebuild / copyback module that issues a next rebuild / copyback request with reference to a progress table that manages the progress of the rebuild / copyback process in response to the completion of the process.

  In the present invention, it is preferable that the RAID control module records the processing completion time of the normal I / O request every time the processing of the normal I / O request is completed, and compares the processing completion time with the current time. Then, it is determined whether a normal I / O request has arrived for a predetermined time, and the rebuild / copy back module inquires of the determination result to the RAID control module, and the request processing size of one time of the rebuild or copy back processing Is set larger than when the normal I / O request arrives for a predetermined time, and the rebuild / copy back request is issued.

  Furthermore, in the present invention, it is preferable that the controller sets the request processing size for one rebuild or copyback process when the normal I / O request does not arrive for a predetermined time. It is set to a multiple of the processing size when the time arrives.

  In the present invention, it is preferable that the controller includes a first interface circuit that performs interface control with the host, a second interface circuit that performs interface control with the plurality of disk devices, and the first interface circuit. An interface circuit connected to the second interface circuit, and having the normal I / O processing and the control unit for executing the rebuild or copyback processing, the control unit receiving an I / O request from the host When a normal I / O request does not arrive for a predetermined time, it is determined whether or not a normal I / O request including the number of times does not arrive for a predetermined time. Is set larger than when the normal I / O request arrives for a predetermined time.

  Further, in the present invention, it is preferable that the controller reads data of a disk device other than the failed disk device when one of the plurality of disk devices constituting the RAID fails, and sends data to the spare disk device. After executing the rebuild process to reconstruct the RAID configuration by performing the operation of writing the data in multiple times, the data of the spare disk device is read, and the new disk provided in place of the failed disk device The operation of writing data to the apparatus is executed in a plurality of times, and copy back processing is executed.

  In the present invention, it is preferable that the control unit has a cache memory for storing a part of the data of the disk device, and the I / O request from the host cannot use the cache memory. An O request is requested as the normal I / O.

In the present invention, the processing size of one rebuild / copyback process is changed according to the presence / absence of normal I / O managed by RAID control. Rebuild / copyback can be executed without hindering the processing, and in a static state where there is no normal I / O, a larger processing size can be specified and the rebuild / copyback processing can be realized at high speed. In addition, the RAID control module that executes normal I / O and rebuild / copy back request at a predetermined ratio and the rebuild / copy back processing module are divided into a few , and the RAID control module determines the presence or absence of normal I / O. Ru can determine and accurately whether the normal I / O load. In addition, even if the rebuild / copyback processing module has normal I / O, a rebuild / copyback request is issued to the RAID control module once , so even if there is normal I / O , rebuild / copyback request at a specified ratio. Even in an environment where the host I / O is less likely to be interrupted, the rebuild / copy back process can be completed by a predetermined time of the rebuild / copy back process with a minimum delay of the host I / O request .

  Hereinafter, embodiments of the present invention will be described in the order of a disk array device (RAID system), thread configuration, rebuild / copy back processing, and other embodiments.

** Disk array device **
FIG. 1 is a configuration diagram of a disk array device according to an embodiment of the present invention, and FIG. 2 is a configuration diagram of a control module of FIG.

  FIG. 1 shows an example of a small-scale disk array device having two control modules, but the same applies to four medium-scale disk array devices and eight large-scale disk array devices.

  As shown in FIG. 1, the disk array device includes a plurality of disk enclosures 2-0 to 2-3 on which a plurality of disk devices for holding data are mounted, a host computer (data processing device) (not shown), and a plurality of disks. Between the enclosures 2-0 to 2-3, there are a plurality (two in this case) of control modules 4-0 to 4-1.

  Each of the control modules 4-0 to 4-1 includes a controller 40, a channel adapter (first interface unit; denoted as CA in the figure) 41, a disk adapter (second interface part; denoted as DA in the figure) 42a, 42 b and a DMA (Direct Memory Access) engine (communication unit; denoted as DMA in the drawing) 43.

  The control modules 4-0 to 4-1 will be described with reference to FIGS. The controller 40 performs read / write processing based on a processing request (read request or write request) from the host computer, and includes a memory 40b and a control unit 40a.

  The memory 40b includes a cache area that holds a part of data held in a plurality of disks in the disk enclosures 2-0 to 2-3, that is, a so-called cache area that serves as a cache for the plurality of disks, and another work area. Have.

  The control unit 40a controls the memory 40b, the channel adapter 41, the device adapter 42, and the DMA 43. For this reason, one or a plurality (two in the figure) of CPUs 400 and 410 and a memory controller 420 are provided. The memory controller 420 controls reading / writing of the memory and switches the path.

  The memory controller 420 is connected to the memory 40b via the memory bus 434, and is connected to the CPUs 400 and 410 via the CPU buses 430 and 432. Further, the memory controller 420 is a 4-lane high-speed serial bus (for example, PCI-Express). The disk adapters 42a and 42b are connected via the 440 and 442, respectively.

  Similarly, the memory controller 420 is connected to the channel adapter 41 (here, four channel adapters 41a, 41b, 41c, and 41d) via a four-lane high-speed serial bus (for example, PCI-Express) 443, 444, 445, and 446. Connected to DMA 43 (here, two DMAs 43-a and 43-b) via a high-speed serial bus (for example, PCI-Express) 447 and 448 of four lanes.

  This high-speed serial bus such as PCI-Express communicates with packets, and by providing a plurality of lanes for the serial bus, even if the number of signal lines is reduced, communication is performed with low delay and high response speed with so-called low latency. be able to.

  The channel adapters 41a to 41d are interfaces to the host computer, and the channel adapters 41a to 41d are connected to different host computers. Each of the channel adapters 41a to 41d is preferably connected to an interface unit of a corresponding host computer by a bus, for example, Fiber Channel or Ethernet (registered trademark). In this case, as the bus, An optical fiber or a coaxial cable is used.

  Further, each of these channel adapters 41a to 41d is configured as a part of each control module 4-0 to 4-1. The channel adapters 41a to 41b support a plurality of protocols as an interface unit between the corresponding host computer and the control modules 4-0 to 4-3.

  Here, since the protocols to be implemented by the corresponding host computers are not the same, they are the main units of the control modules 4-0 to 4-1 so that the channel adapters 41a to 41d can be easily replaced as necessary. The controller 40 is mounted on a separate printed circuit board.

  For example, as described above, the protocol between the host computers to be supported by the channel adapters 41a to 41d includes Fiber Channel, iSCSI (Internet Small Computer System Interface) corresponding to Ethernet (registered trademark), and the like.

  Further, as described above, each of the channel adapters 41a to 41d is directly coupled to the controller 40 by a bus designed for connecting between an LSI (Large Scale Integration) and a printed circuit board, such as a PCI-Express bus. Has been. Thereby, the high throughput requested | required between each channel adapter 41a-41d and the controller 40 is realizable.

  The disk adapters 42a and 42b are interfaces to the respective disk drives of the disk enclosures 2-0 to 2-3, and are connected to the disk enclosures 2-0 to 2-3. Here, four FC (Fiber Channel) ports are connected. Have.

  Further, as described above, each of the disk adapters 42a and 42b is directly coupled to the controller 40 by a bus designed to connect between an LSI (Large Scale Integration) and a printed circuit board, such as a PCI-Express bus. Has been. As a result, a high throughput required between the disk adapters 42a and 42b and the controller 40 can be realized.

  As shown in FIG. 1, each of the disk enclosures 2-0 to 2-3 has a plurality of (for example, 15) disk drives 200 each having two ports. In the disk enclosures 2-0 to 2-3, each port of each disk drive 200 is connected to two ports by a pair of FC cables from the two ports.

  These two ports connect the disk adapters 42a and 42b of the respective control modules 4-0 to 4-1 to all the disk enclosures 2-0 to 2-3. That is, the disk adapter 42a of each control module 4-0 to 4-1 is connected to the disk enclosures 2-0 to 2-3. Similarly, the disk adapters 42b of the control modules 4-0 to 4-1 are connected to the disk enclosures 2-0 to 2-3.

  With such a configuration, the control modules 4-0 to 4-1 are connected to all the disk enclosures (disk drives) 2-0 to 2-3 through any of the disk adapters 42a and 42b and through any path. Accessible.

  The disk adapters 42a and 42b and the corresponding disk enclosures 2-0 to 2-3 are connected by, for example, fiber channels. In this case, since the modules are different, they are connected by optical cables.

  As shown in FIG. 1, the DMA engine 43 communicates with the control modules 4-0 to 4-1, and is responsible for communication and data transfer processing between other control modules. Each of the DMA engines 43 of the control modules 4-0 to 4-1 is configured as a part of the control modules 4-0 to 4-1, and is a main unit of the control modules 4-0 to 4-1. It is mounted on the substrate of the controller 40. Then, it is directly coupled to the controller 40 by the above-described high-speed serial bus and communicates with the DMA engines 43 of the other control modules 4-0 to 4-1.

  With such a configuration, each of the DMA engines 43 of the control modules 4-0 to 4-1 can be connected between the controller 40 connected thereto and the controllers 40 of the other control modules 4-0 to 4-1. Communication or data transfer processing (for example, mirroring processing) that occurs in response to an access request from the host computer is executed.

  As shown in FIG. 2, the DMA engines 43 of the control modules 4-0 to 4-7 are composed of a plurality of (here, two) DMA engines 43-a and 43-b, and these two DMA engines. As shown in FIG. 2, each of 43-a and 43-b is connected to the controller 40 by, for example, a PCI-Express bus. That is, in communication and data transfer (DMA) processing between the control modules 4-0 to 4-1 (that is, between the controllers 40 of the control modules 4-0 to 4-1), the amount of data transfer is large and communication control is performed. It is desirable to shorten the time required for the processing, and high latency and low latency (fast response speed) are required. For this reason, as shown in FIGS. 1 and 2, the DMA engine 43 of each control module 4-0 to 4-1 performs high-speed serial transmission designed to satisfy both requirements of high throughput and low latency. Connected by the used bus (PCI-Express or Rapid-IO).

  These PCI-Express and Rapid-IO use high-speed serial transmission of 2.5 Gbps, and a low-amplitude differential interface called LVDS (Low Voltage Differential Signaling) is adopted for these bus interfaces.

** Thread configuration **
Next, software program modules (referred to as threads) executed by the CPU 400 (410) will be described. FIG. 3 is a configuration diagram of software threads executed by the CPU 400 (410), and FIG. 4 is a partial detail view thereof. As shown in FIG. 3, the software module includes a kernel / scheduler 600, an IO control module 602, an inter-CM communication driver 604, an inter-CPU communication driver 606, a system control module 608, and a network driver 610.

  The kernel / scheduler 600 is an operating system such as MS-DOS (registered trademark). The IO control module 602 includes a CM-CA driver 620, a basic module 640, and a CM-DA driver 630. The CM-CA driver 620 is a driver that drives a CA (channel adapter) 41.

  The basic module 630 includes a resource thread (resource control module) 646 that performs resource management, a copy thread (copy control module) 646 that performs copy processing, and a cache thread (cache memory control module) 648 that controls the cache memory of the memory 40b. And a back end having a RAID thread (RAID control module) 632 for performing RAID configuration control and an OVSM thread (OVSM control module) 634.

  The OVSM thread 634 performs a rebuild / copy back process to be described later. The maintenance agent 612 makes various notifications to the OVSM thread 634. The CM-DA driver 630 is a driver that drives a DA (disk adapter) 42. The inter-CM communication driver 604 communicates with other CMs, and the inter-CPU communication driver 606 communicates with other CPUs 410.

  The system control module 608 has a configuration management module 652. The configuration management module 652 creates a configuration definition table including an RLU (Raid Logical Unit) table for configuration definition in the memory 40b in accordance with designation from the service controller 46 via the network driver 610. The service controller 46 is not shown in FIG. 1 and is common to each CM.

  The disk drive 200 provided in the disk enclosures 2-0 to 2-3 connected to the DA 42 is divided into a system disk 200-1 and a user disk 200-2. A table is also stored.

  The resource module 642 and the RAID module 632 refer to the configuration definition table, convert a LUN (logical unit number) from the host into a PLBA (physical block address) of the physical disk, and access the physical disk 200.

  As illustrated in FIG. 4, the OVSM thread 634 includes a Rebuild / Copyback module 634-1 that manages the progress of Rebuild / Copyback. The Rebuild / Copyback module 634-1 requests the RAID control module 632, which will be described later, to process the Rebuild / Copyback once, obtains a response, and records the progress of the Rebuild / Copyback.

  On the other hand, the RAID control module 632 acquires a buffer for temporarily saving data (hereinafter referred to as “Data Buffer”), and instructs the disk device to read or write. In particular, Rebuild / Copyback is in charge of processing to read from the disk device, store it in the Data Buffer, and write the contents of the Data Buffer to the disk device.

  The RAID control module 632 includes a normal I / O completion time recording unit 632-1 that records the time at which the last normal I / O is completed, a time of the normal I / O completion time recording unit 632-1, and the current For example, a normal I / O presence / absence determination module 632-2 that compares the time and determines whether there is a difference of 5 seconds, for example.

  The OVSM module 634 also includes the Rebuild / Copyback modules 634-2 and 634-3 that request the Rebuild / Copyback actual processing to the RAID control module 632, and the normal I / O presence / absence determination module 632-2 at the time of actual processing request. A processing size change module 634-4 that calls the result and changes the processing size of one actual processing according to the result is provided.

  The Rebuild / Copyback module 634-1 determines a single actual processing size by referring to the progress table 410 storing the start LBA provided in the memory 40b and the default storage unit 414 storing the default value of the processing size. And request actual processing.

  With such a configuration, each time the OVSM module 634 requests the RAID control module 632 to execute the actual processing of Rebuild / Copyback, the OVSM module 634 makes a normal I / O presence / absence to the RAID control module 632 (normally The OVSM module 634 receives the result and confirms whether the Rebuild / Copyback actual processing parameters to be passed to the RAID control module 632 (once) Change the processing size).

  Here, the determination of the presence / absence of normal I / O is set to 5 sec or more in order to prevent misidentification when the timing with and without I / O is repeated alternately. In other words, if sparsely normal I / O is coming, even if it appears that there is no normal I / O at the timing of Rebuild / Copyback, it will be normal immediately after starting Rebuild / Copyback. I / O may occur. Such a case means that the normal I / O is present and the normal I / O is not kept waiting.

** Rebuild / Copyback process **
FIG. 5 is a flowchart of rebuild / copyback processing according to the embodiment of the present invention, FIG. 6 is a time chart thereof, and FIG. 7 is an operation explanatory diagram thereof.

  The rebuild / copy back process of FIG. 5 will be described below with reference to FIGS.

  (S10) First, the OVSM module 634 confirms that there is a completion response from the RAID control module 632 at the start of rebuild / copyback.

  (S12) Next, the rebuild / copy back module 634-1 of the OVSM module 634 refers to the start RLBA in the rebuild / copy back progress table 410 of the memory 40b. As shown in FIG. 7, when the target of the rebuild / copy back process is LBA (Logical Block Address) 000 to nnn in the RAID configuration, the start RLBA includes the start RLBA of the target data of the next rebuild / copy back process. Is recorded.

  (S14) The rebuild / copy back module 634-1 determines from the start RLBA of the progress table 410 whether processing of all target RLUs (Raid Logical Units) has been completed. When the rebuild / copy back module 634-1 determines that all the RLUs have been processed, the rebuild / copy back module 634-1 ends.

  (S16) Conversely, if the rebuild / copy back module 634-1 determines that the processing of all the RLUs has not been completed, the rebuild / copy back module 634-1 inquires of the RAID control module 632 about the presence of normal I / O. As described with reference to FIG. 4, the RAID control module 632 determines whether or not 5 seconds or more have elapsed since the normal I / O was absent.

  (S18) The rebuild / copy back module 634-1 confirms the presence / absence of normal I / O from the response from the RAID control module 632, and if there is normal I / O, it is shown as copy back 1 in FIG. As described above, the one-time processing size is set to the default value of the memory 40b. On the other hand, when there is no normal I / O, as shown in copy back 2 of FIG.

  (S20) Then, the rebuild / copy back module 634-1 requests the RAID control module 632 for actual processing having the processing size set as the start RLBA. Further, the rebuild / copy back module 634-1 adds the processing size to the start RLBA, calculates the next start RLBA, and updates the progress table 410 to the calculated start RLBA. Then, the process returns to step S10.

  In this way, in the case of static with the state with normal I / O (hereinafter referred to as “dynamic”) and the state with no I / O (hereinafter referred to as “static”), the processing size for one time is increased. To do. If the size of one time is large, the Rebuild / Copyback process becomes faster, but the influence on the normal I / O becomes larger instead.

  For this reason, in the dynamic case, a processing size that does not interfere with normal I / O processing is specified, and in the static case, a larger processing size (for example, a size of one track or more of a physical disk drive) Is desirable in terms of disk performance). In order to specify this size, it is only necessary to change the parameter when the Rebuild / Copyback actual processing request is sent from the OVSM 634 to the RAID control 632.

  3, when the CA 41 receives a read request from the host, the CA 41 notifies the resource module 642 via the CM-CA driver 620, and the resource module 642 Accept a read request.

  The resource module 642 requests the cache module 648 to determine whether or not there is a cache hit. The cache module 648 checks whether the LBA (OLBA) data of the target host LUN exists in the cache area of the memory 40b, and if it exists (hit), the OLBA (host logical block address) data of the memory 40b. And the data is transferred from the CA 41 to the host via the CM-CA driver 620.

  On the other hand, if the cache module 648 determines that the LBA (OLBA) data of the target host LUN does not exist in the cache area of the memory 40b (mis-hit), the cache module 648 sends the back-end RAID control module 632 from the physical disk. Make a data read request. As is well known, the RAID module 632 converts the LBA (OLBA) of the requested host LUN into the LBA (RLBA) of the RAID group.

  Next, the RAID control module 632 converts the LBA (RLBA) of the RAID group into the LBA (DLBA) of the virtual disk. Next, the RAID module 632 converts the LBA (DLBA) of the virtual disk requested to be processed into the LBA (PLBA) of the physical disk. This process differs depending on the RAID configuration (for example, RAID 1 and RAID 5).

  Next, the RAID module 632 obtains a disk enclosure RAID and slot from the LBA (PLBA) of the physical disk requested to be processed, and makes a read request to the LBA (PLBA) of the physical disk 200. That is, a read request for the physical disk is issued to the corresponding DA 42 via the CM-DA driver 630, and the physical disk is accessed via the BRT 5-0.

  When data is read from the physical disk, the cache module 648 stores it in the cache area of the memory 40b via the DA 42, and when the read is completed, the resource module 642 sends this read data from the CA 41 to the host via the CM-CA driver 620. To exit.

  In the write process from the host, the write data is temporarily stored in the cache area of the memory 40b and then written back to the physical disk according to the internal write-back schedule. Similarly, at the time of this write back, the address conversion processing using the configuration definition table is performed for the processing after the miss hit.

  Also, the RAID control module 632 gives priority to the received host I / O, normal I / O such as internal I / O such as write back, and the received rebuild / copy back request at a predetermined I / O ratio. And request to the CM-DA module 630 of the lower layer in order of priority.

  In this way, the processing size of one rebuild / copyback process is changed according to the presence / absence of normal I / O managed by RAID control. Therefore, when there is normal I / O, normal I / O is present. Therefore, rebuild / copyback processing can be executed, and in a static state, a larger processing size can be specified, and rebuild / copyback processing can be realized at high speed.

  The RAID control module 632 records the time when the last normal I / O is completed, compares this time with the current time, and determines whether there is a difference of 5 seconds, for example. Since the O presence / absence determination process is performed, the rebuild / copy back module 634-1 can easily change the process size and proceed with the rebuild / copy back process by an inquiry.

** Other embodiments **
In the embodiment described above, the disk array device having the configuration as shown in FIG. 1 has been described, but the present invention can be applied to a disk array device having a configuration other than this. As the physical disk, a magnetic disk, an optical disk, a magneto-optical disk, and various storage devices can be applied.

  Furthermore, although the monitoring time for the presence / absence of normal I / O is set to 5 seconds, other monitoring time may be used, and the determination of the presence / absence of I / O can also be executed by the OVSM module. In addition, although the processing size is changed to twice the default value, other multiples such as 1.5 times can be used, and the processing size may be provided separately depending on whether or not normal I / O is present.

  As mentioned above, although this invention was demonstrated by embodiment, in the range of the meaning of this invention, this invention can be variously deformed, These are not excluded from the scope of the present invention.

  (Supplementary Note 1) In a RAID system that accesses a connected disk device in response to an I / O request requested from a host and executes the I / O request, a plurality of the disk devices constituting a RAID, and the plurality of disk devices When one of the disk devices fails, the data of the disk device other than the failed disk device is read, and the operation of writing the data to the spare disk device or the new disk device is executed in multiple times to create a RAID configuration. And a controller for executing a rebuild or copyback process for reconstructing the I / O, and the controller determines whether or not a normal I / O request including an I / O request from the host comes for a predetermined time, and The request processing size of one rebuild or copyback process executed separately is set to the above when a normal I / O request does not arrive for a predetermined time. RAID system, characterized in that the normally I / O requests than arriving predetermined period was increased.

  (Supplementary Note 2) The controller records the processing completion time of the normal I / O request every time the processing of the normal I / O request is completed, compares the processing completion time with the current time, and performs normal I / O The RAID system according to appendix 1, wherein it is determined whether the request does not come for a predetermined time.

  (Supplementary Note 3) The controller balances and executes the normal I / O request and the one request of the rebuild / copyback, and performs the following in response to the completion of the processing of the rebuild / copyback request. The RAID system according to appendix 1, wherein a rebuild / copy back request is issued.

  (Supplementary Note 4) The controller has a progress table for managing the progress status of the rebuild / copyback process, creates one request for rebuild or copyback process according to the progress table, and stores the progress table. The RAID system according to Appendix 1, wherein the RAID system is updated.

  (Supplementary Note 5) The controller balances the normal I / O request and the rebuild / copy back request once, and executes the RAID control module to be executed and the completion of the rebuild / copy back request processing. The RAID system according to appendix 3, further comprising: a rebuild / copyback module that issues a next rebuild / copyback request with reference to a progress table that manages a progress status of the rebuild / copyback processing.

  (Supplementary Note 6) The RAID control module records the processing completion time of the normal I / O request every time the processing of the normal I / O request is completed, compares the processing completion time with the current time, The rebuild / copy back module inquires of the RAID control module about the determination result, and determines the request processing size for one rebuild or copy back process as a normal I / O. The RAID system according to appendix 5, wherein when the O request does not arrive for a predetermined time, the rebuild / copy back request is issued with a larger setting than when the normal I / O request has arrived for the predetermined time.

  (Supplementary note 7) The controller sets the request processing size of the rebuild or copyback process for one time when the normal I / O request does not arrive for a predetermined time and the normal I / O request arrives for a predetermined time. The RAID system according to Supplementary Note 1, wherein the RAID system is set to a multiple of the processing size.

  (Supplementary Note 8) The controller includes a first interface circuit that performs interface control with the host, a second interface circuit that performs interface control with the plurality of disk devices, the first interface circuit, and the first interface circuit. The normal I / O process and the control unit that executes the rebuild or copyback process, and the control unit includes a normal I / O request including an I / O request from the host. It is determined whether the O request does not arrive for a predetermined time, and the request processing size of one rebuild or copyback process executed separately for a plurality of times is set as the normal I / O request when the normal I / O request does not arrive for the predetermined time. The RAID system according to appendix 1, wherein the / O request is set larger than when the predetermined time has arrived.

  (Supplementary Note 9) When one of the plurality of disk devices constituting the RAID fails, the controller reads data from disk devices other than the failed disk device and writes data to the spare disk device. After executing the rebuild process to reconstruct the RAID configuration by executing multiple times, read the data of the spare disk device and write the data to the new disk device provided in place of the failed disk device The RAID system according to appendix 1, wherein the copy back processing is executed by performing the operation to be performed a plurality of times.

  (Supplementary Note 10) The control unit includes a cache memory for storing a part of the data of the disk device, and among the I / O requests from the host, the I / O requests that cannot use the cache memory are The RAID system according to appendix 5, characterized in that it is requested as normal I / O.

  (Supplementary Note 11) In a rebuild / copyback processing method of a RAID system in which a connected disk device is accessed and an I / O request is executed in response to an I / O request requested from a host, a plurality of components constituting a RAID When one of the disk devices fails, the data of the disk device other than the failed disk device is read, and the operation of writing the data to the spare disk device or the new disk device is executed in multiple times, and the RAID configuration Executing a rebuild or copyback process for reconfiguring, a step for determining whether a normal I / O request including an I / O request from the host does not arrive for a predetermined time, and a rebuild performed separately for a plurality of times Alternatively, if the normal I / O request does not arrive for a predetermined time, the normal I / O request processing size for one copy back process is set. Than if the determined arrives predetermined time, rebuild / copy back processing method of the RAID system, characterized by a step of larger.

  (Supplementary Note 12) In the determination step, the process completion time of the normal I / O request is recorded every time the process of the normal I / O request is completed, and the process completion time is compared with the current time to calculate the normal I / O request. The rebuild / copyback processing method for a RAID system according to appendix 11, characterized by comprising a step of determining whether or not the O request has come for a predetermined time.

  (Supplementary Note 13) The method further includes a step of balancing and executing the normal I / O request and the one request of rebuild / copy back, and the execution step of the rebuild / copy back process includes the rebuild / copy back The rebuild / copyback processing method for a RAID system according to appendix 11, further comprising a step of issuing a next rebuild / copyback request in response to completion of processing of the back request.

  (Additional remark 14) The execution step of the rebuild / copy back process refers to a progress table that manages the progress status of the rebuild / copy back, and creates a single request for the rebuild or copy back process; The rebuild / copyback processing method for a RAID system according to appendix 11, further comprising a step of updating the progress table.

  (Supplementary Note 15) The RAID control module further includes a step of balancing and executing the normal I / O request and the one request of rebuild / copy back, and the processing step includes the rebuild / copy back. The method includes a step of issuing a next rebuild / copy back request to the RAID control module with reference to a progress table for managing a progress status of the rebuild / copy back processing in response to completion of processing of the request. The rebuild / copyback processing method of the RAID system according to appendix 11.

  (Supplementary Note 16) In the determination step, the RAID control module records the process completion time of the normal I / O request every time the process of the normal I / O request is completed, and compares the process completion time with the current time. The rebuild / copyback request issuance step inquires of the RAID control module about the determination result and determines whether the rebuild / copyback process is one of the rebuild or copyback processes. A step of issuing the rebuild / copyback request by setting the request processing size of the number of times larger than when the normal I / O request arrives for a predetermined time if the normal I / O request does not arrive for a predetermined time The rebuild / copyback processing method for a RAID system according to appendix 15, characterized by comprising:

  (Supplementary Note 17) In the setting step, when the normal I / O request does not arrive for a predetermined time if the normal I / O request does not arrive for a predetermined time, The rebuild / copyback processing method for a RAID system according to appendix 11, characterized by comprising a step of setting a multiple of the processing size.

  (Supplementary Note 18) The determination step is connected to a first interface circuit that performs interface control with the host and a second interface circuit that performs interface control with the plurality of disk devices, and the normal I / O Processing and a step of determining whether a normal I / O request including an I / O request from the host does not arrive for a predetermined time by the control unit that executes the rebuild or copyback processing, and the setting step includes If the normal I / O request does not arrive for a predetermined time, the normal I / O request has arrived for a predetermined time when the normal I / O request does not arrive for a predetermined time for the rebuild or copyback process executed separately by the control unit. The rebuild / copy back of the RAID system according to appendix 11, characterized by comprising a step of setting larger than the case. Processing method.

  (Supplementary Note 19) The step of executing the rebuild / copy back process reads data of a disk device other than the failed disk device when one of the plurality of disk devices constituting the RAID fails, A rebuild process step for rebuilding a RAID configuration by performing an operation of writing data to the apparatus in a plurality of times, and after executing the rebuild process, the data of the spare disk device is read and the failed disk The rebuild / copyback of the RAID system according to appendix 11, characterized by comprising a step of executing a copyback process by performing an operation of writing data to a new disk device provided instead of the device in a plurality of times Processing method.

  (Additional remark 20) It further has the step which requests | requires the I / O request which cannot use the cache memory which stores a part of data of the said disk unit among the I / O request from the said host as said normal I / O. The rebuild / copy back processing method of the RAID system according to appendix 15, wherein

As described above, the processing size for one rebuild / copyback process is changed according to the presence / absence of normal I / O managed by RAID control. Rebuild / copyback can be executed without hindering / O processing, and in a static state where there is no normal I / O, a larger processing size can be specified to achieve rebuild / copyback processing at high speed. it can. In addition, the RAID control module that executes normal I / O and rebuild / copy back request at a predetermined ratio and the rebuild / copy back processing module are divided into a few, and the RAID control module determines the presence or absence of normal I / O. The presence / absence of normal I / O can be accurately determined by load. In addition, even if the rebuild / copy back processing module has normal I / O, a rebuild / copy back request is issued to the RAID control module once, so even if there is normal I / O, a rebuild / copy back request is made at a predetermined ratio. Even in an environment where the host I / O is less likely to be interrupted, the rebuild / copy back process can be completed by a predetermined time of the rebuild / copy back process with a minimum delay of the host I / O request.

It is a block diagram of the disk array apparatus of one embodiment of this invention. It is a block diagram of the control module of FIG. FIG. 3 is a functional block diagram of firmware of the control module in FIGS. 1 and 2. FIG. 4 is a detailed configuration diagram of the firmware in FIG. 3. It is a rebuild / copy back process flowchart of one embodiment of this invention. It is a time chart figure of the processing of FIG. It is operation | movement explanatory drawing of the process of FIG. It is explanatory drawing of the conventional rebuild process. It is explanatory drawing of the conventional copyback process.

Explanation of symbols

4-0 to 4-1 Control module 40 Control unit 41 Channel adapter 42a, 42b Device adapter 400, 410 CPU
40b Memory 200 Disk device 632 RAID control module 634-1 Rebuild / copy back module 2-0 to 2-3 Device enclosure

Claims (5)

In a RAID system that accesses a connected disk device and executes an I / O request in response to an I / O request requested by a host,
A plurality of the disk devices constituting a RAID;
When one of the plurality of disk devices fails, the data of the disk device other than the failed disk device is read, and the operation of writing the data to the spare disk device or the new disk device is executed in multiple times, A controller that executes rebuild or copyback processing for reconstructing the RAID configuration,
The controller is
A RAID control module that prioritizes the normal I / O request and the rebuild / copyback request once and executes the request at a predetermined ratio;
A rebuild / copyback module that issues a next rebuild / copyback request with reference to a progress table that manages the progress of the rebuild / copyback processing in response to the completion of the rebuild / copyback request processing. And
The RAID control module records the processing completion time of the normal I / O request every time processing of the normal I / O request is completed, and compares the processing completion time with the current time to determine whether the normal I / O request is Determine if it will come within the specified time,
The rebuild / copy back module inquires the determination result to the RAID control module in response to the completion of the rebuild / copy back request processing , and sets the request processing size of the rebuild or copy back processing once to the normal I / If the O request does not arrive within a predetermined time, from when the normal I / O request has arrived within a predetermined period of time, is set larger, once the request of the rebuild / copy back, the RAID control module A RAID system characterized by being issued to. The controller has a cache memory for storing a part of the data of the disk device,
2. The RAID system according to claim 1, wherein an I / O request that cannot use the cache memory is requested from the host to the RAID control module as the normal I / O request. The rebuild / copy back module sets the request processing size for one rebuild or copy back process within a predetermined time when the normal I / O request does not arrive within a predetermined time. The RAID system according to claim 1, wherein the RAID system is set to a multiple of the processing size when it arrives. In a rebuild / copyback processing method of a RAID system in which a controller accesses a connected disk device in response to an I / O request requested from a host and executes the I / O request.
When one of the plurality of disk devices constituting the RAID fails, the operation of reading the data of the disk device other than the failed disk device and writing the data to the spare disk device or the new disk device is divided into multiple times. Performing a rebuild or copyback process to reconstruct the RAID configuration; and
The RAID control module records the processing completion time of the normal I / O request every time the processing of the normal I / O request is completed, compares the processing completion time with the current time, and receives the I / O from the host. Determining whether a normal I / O request including the request does not come within a predetermined time;
A step of executing the normal I / O request and the one rebuild / copy back request at a predetermined ratio according to priority by a RAID control module;
The execution step of the rebuild / copyback process includes:
When the rebuild / copy back request is completed, the rebuild / copy back module refers to the progress table that manages the progress of the rebuild / copy back process and starts generating the next rebuild / copy back request. Steps,
The rebuild / copy back module inquires the determination result to the RAID control module upon completion of processing of the rebuild / copy back request , and sets the request processing size of one time of the rebuild or copy back processing to the normal I / If the O request does not arrive within a predetermined time, from when the normal I / O request has arrived within a predetermined period of time, is set larger, once the request of the rebuild / copy back, the RAID control module A rebuild / copyback processing method for a RAID system. The controller requests, as the normal I / O request, the RAID control module for an I / O request that cannot use the cache memory for storing a part of the data of the disk device among the I / O requests from the host. The RAID system rebuild / copyback processing method according to claim 4, further comprising: a step of:

Images